As described in the introductory presentation that I posted last month, my first main activity has been to improve the system integration modeling in WITCH, in order to align our model with the state of the art of Integrated Assessment Models as resulted from the ADVANCE project. After the preliminary literature review that I had conducted in the very first weeks of my project, I lately studied more in depth the different solutions developed by the other models and I eventually decided to follow the path of the MESSAGE model. This is the model that had first implemented the modeling mechanism based on the flexibility and the capacity constraints, which are the core of system integration modeling in WITCH as well. The new modeling in MESSAGE does not directly implement the Residual Load Duration Curves (RLDCs) developed during the ADVANCE project by our research partners, but rather it indirectly uses them to refine the formulation of the abovementioned constraints, while keeping on treating electricity as a homogeneous good over the year. I considered that this is a very effective way to implement the information richness included in the RLDCs, assuring at the same time modeling simplicity and manageability.
In particular, I:
- added curtailment of VRE overproduction
- updated the capacity constraint (firm capacity requirement and capacity value of wind and solar PV)
- updated the flexibility constraint (operating reserve coefficients for VRE and load and representation of flexible thermoelectric plant operation)
I also revised the modeling of plant depreciation, describing more accurately the ageing of power plants.
Naturally, the modeling details will be thoroughly described in the first project deliverable which is due in Autumn (D1.1). Before that, however, I will have the chance of presenting these topics in the internal seminar that I have planned to give at FEEM during my “European break” in June.
In these weeks I carried out the preliminary work, i.e. I identified the equations to implement in the model and developed the supporting Excel files, deriving data for the WITCH regions. Now I am ready to proceed with the actual implementation in the code, which normally is not a “difficult” activity from a scientific point of view, but might be complicated and annoying from a practical point of view, since infeasibilities and computational errors always lie in ambush 🙂